INDEX

Explanations

phrases indicating temporal relationships or connections

oai_token-act-pair · gpt-4o-mini Triggered by @bot

independently or given

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

featureID

-0.91

 Roskov

-0.82

verwijspagina

-0.71

 disambiguazione

-0.69

StructEnd

-0.63

InitVars

-0.61

 فريبيس

-0.61

 defStyleAttr

-0.60

AndEndTag

-0.60

 autorytatywna

-0.59

POSITIVE LOGITS

 equally

0.35

ENOT

0.35

 igualmente

0.33

yet

0.32

tvguidetime

0.32

 similarly

0.31

 necesariamente

0.31

Fä

0.31

 inom

0.30

 necessarily

0.30

Activations Density 0.153%

phrases indicating temporal relationships or connections

independently or given

No Comments

No Known Activations